Implementation method of omni-channel power IT service self-service response

Abstract

The invention relates to an implementation method of omni-channel power IT service self-service response. The implementation method comprises the following steps: establishing a response database; obtaining a query habit and query content of the user in a questionnaire form; initial reply information and query guidance are provided for the user according to the reply level; acquiring subsequent reply information corresponding to the query guidance; the manual service platform gives ideal reply information; the manual service platform compares the ideal reply information with reference information to obtain difference reply content; determining a modification level of the differential reply content; and modifying the reference information according to the modification level to obtain modification reply information, replacing the reference information with the modification reply information, and inputting the reference information into a reply database. According to the implementation method, the response database is established, dynamic adjustment is carried out according to the query habit and content of the user, follow-up operation of the user is tracked, feedback is obtained, the content in the response database is continuously learned and optimized, intelligent iteration is achieved, and the system is more perfect.

Description

Technical Field

[0001] The present invention relates to the field of IT service technology, and in particular to a method for implementing self-service response of omni-channel power IT services. Background Art

[0002] With the development of the times, IT service work currently has problems such as complicated systems, large maintenance of business process documents and system operation documents, and untimely updates. The operation methods of various business systems are different. Relying solely on training staff to complete business management will result in excessive human cost investment.

[0003] "CN111192060B" discloses a method for implementing omni-channel self-service answering based on power IT services, which handles various businesses through human-computer collaboration. However, when reviewing questions raised by users, the questioning habits of different users are not fully taken into consideration. If the system is only modified from the perspective of providing answers, the system will take too long to improve. During the system improvement process, the user experience is also poor, and the system cannot form a good learning method for the user's personal habits, which leads to slower system updates and iterations. Therefore, how to provide users with reasonable replies and guidance based on their query habits is one of the problems that technical personnel in this field urgently need to solve. Summary of the invention

[0004] The problem solved by the present invention is: how to provide reasonable answers and guidance to users according to their query habits.

[0005] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0006] A method for implementing omni-channel power IT service self-service response, comprising the following steps:

[0007] S100, establishing a response database, and classifying response levels according to response keywords;

[0008] S200, obtaining the user's query habits and query content through a questionnaire, and adjusting them to obtain adjusted query content;

[0009] S300, determining a response level for adjusting the query content, and providing the user with initial response information and query guidance according to the response level;

[0010] S400, tracking the user's subsequent operations and obtaining subsequent response information corresponding to the query guidance;

[0011] S500: When the number of guidance times reaches the guidance threshold, the manual service platform gives an ideal response information;

[0012] S600, the initial reply information and the subsequent reply information are recorded as reference information, and the manual service platform compares the ideal reply information with the reference information to obtain differential reply content;

[0013] S700, determining the difference keywords that cause the difference reply content, obtaining the response level and the frequency of occurrence of the difference keywords in the reference information, and determining the modification level of the difference reply content;

[0014] S800, modify the reference information according to the modification level to obtain modified reply information, replace the reference information with the modified reply information, and input it into the response database.

[0015] Preferably, in step S200, obtaining the adjusted query content includes the following steps:

[0016] S210, performing word order detection on the questionnaire survey results, obtaining the word order habits of the user when performing a query, and obtaining a word order adjustment template;

[0017] S220, obtaining the user's replacement habit for synonyms when performing a query through vocabulary detection, and obtaining a vocabulary adjustment template;

[0018] S230: Adjust the initial query content according to the vocabulary adjustment template and the word order adjustment template to obtain adjusted query content.

[0019] Preferably, in step S300, the user's historical query records are obtained, the response level for adjusting the query content is determined according to the historical query records, and initial reply information and query guidance are provided according to the response level.

[0020] Preferably, the method comprises the following steps:

[0021] S310, obtaining historical keywords of the historical query record, matching the historical keywords with the response content in the response database, screening the response content according to the matching similarity, and obtaining the judgment content;

[0022] S320, screening out the minimum value of the response level corresponding to the determination content, and determining the response level of the adjusted query content according to the minimum value of the response level;

[0023] S330, determining the detail of the response content according to the response level to obtain initial reply information, and supplementing the query guidance according to the determined content.

[0024] Preferably, in step S600, obtaining the difference reply content includes the following steps:

[0025] S610, summarizing and integrating the content of the ideal reply information to obtain ideal keywords;

[0026] S620, recording the initial reply information or the subsequent reply information that does not contain the ideal keyword in the reference information as an erroneous reply content;

[0027] S630: Record the initial reply information or the subsequent reply information containing the ideal keyword in the reference information as differential reply content.

[0028] Preferably, in step S700, the difference reply contents are summarized and integrated to obtain difference keywords, the response level and the occurrence frequency of the difference keywords in the reference information are obtained, and the modification level of the difference reply contents is determined, including the following steps:

[0029] S710, obtaining the number of occurrences of the difference keywords in each difference reply content, and calculating the frequency of occurrence of the difference reply content according to the number of occurrences of the difference keywords;

[0030] S720, combining the occurrence frequency and the response level of each difference reply content to obtain a modification coefficient of each difference reply content;

[0031] S730. Determine the modification level of the difference reply content according to the modification coefficient.

[0032] Preferably, in step S720, obtaining the modification coefficient comprises the following steps:

[0033] S721, calculating the difference percentage corresponding to each response level according to the occurrence frequency, and comparing the difference percentage with the difference threshold;

[0034] S722, when the difference percentage is greater than or equal to the difference threshold, adjusting the response level;

[0035] S723. When the difference percentage is less than the difference threshold, calculate a modification coefficient according to the difference percentage.

[0036] Preferably, in step S710, the number of words in the difference reply content is recorded as A1, the number of words in the difference keyword is recorded as A2, the number of occurrences of the difference keyword is recorded as B, the frequency of occurrence is recorded as Z, and the following relationship is satisfied among A1, A2, B and Z:

[0037] Z = A1÷(A2×B);

[0038] In step S720, different response levels have different modification coefficients. The larger the response level, the larger the corresponding modification coefficient. Let the response level be L1, the modification coefficient be F1, and the difference percentage be F0. Then, the following relationship is satisfied among F0, F1 and L1:

[0039] F1=1+F0×(0.8+0.4L1).

[0040] Preferably, the step S800 includes the following steps:

[0041] S810, finding corresponding subsequent reply information from the reference information according to the modification level; when the modification level is the same as the response level of the subsequent reply information, adding the ideal reply information to the subsequent reply information to obtain modified reply information;

[0042] S820, when the modification level is different from the response level of the subsequent reply information, calculating the level difference between the modification level and the response level, and recording the subsequent reply information between the level differences as information to be adjusted;

[0043] S830, integrating the information to be adjusted and the ideal response information to obtain multiple response vectors, performing similarity analysis on the multiple response vectors and the initial response information, and determining to modify the response information according to the analysis result.

[0044] Preferably, the step S830 includes the following steps:

[0045] S831, filter out a reply vector containing ideal reply information and record it as a vector to be replaced;

[0046] S832: Compare the similarities of the vectors to be replaced, use the vector to be replaced with the highest similarity as the modified reply information, and adjust the query guidance according to the order of similarity from high to low.

[0047] The method for implementing the self-service response of an omni-channel power IT service of the present invention has the following beneficial effects:

[0048] By establishing an answer database and dynamically adjusting it according to the user's query habits and content, in-depth analysis and optimization of user query behavior can more accurately guide users to find the information they need, reduce invalid operations during the query process, and thus improve the efficiency of self-service; by tracking the user's subsequent operations and obtaining feedback, when necessary, the ideal answer information given by the manual service platform is compared with the reference information, and the content in the answer database is continuously learned and optimized to achieve intelligent iteration and make the answer database more complete; by analyzing the difference in answer content, the modification level is determined, and the reference information is modified accordingly. This refined management method can ensure that the content quality of the answer database continues to improve, and provide users with more accurate and comprehensive answers.

[0049] The setting of the response database allows the self-service response system to store enough response keywords. The division of response levels can provide corresponding reply information based on the user's query information. When the user enters fewer response keywords, reasonable query guidance can be provided to the user to improve the user's query experience. The acquisition of query habits can formulate different keyword reading methods according to the habits of different users, which improves the practicality of the self-service response method. By tracking the subsequent query content, it can determine the user's satisfaction with the initial reply information, and determine whether the initial reply information needs to be modified based on the satisfaction level. By setting the number of guidance times and guidance thresholds, the manual service platform can intervene in time when the self-service response system cannot provide satisfactory reply information. By obtaining the response level and frequency of occurrence of different keywords, it can determine the reference information that needs to be replaced for the different reply content. After the reference information is modified by the different reply content, the initial reply information corresponding to each response level is more reasonable.

[0050] The setting of the word order adjustment template can upgrade the self-service answering system according to the speech habits of different users, and can more accurately identify the initial query content entered by the user through voice input. The setting of the vocabulary adjustment template allows the self-service answering system to actively adjust the initial query content entered by the user according to the user's word usage habits, so that the self-service answering system can provide appropriate initial reply information.

[0051] The acquisition of historical keywords can obtain the user's query content before using the self-service answering system, and can quickly understand the user's understanding of the answer keywords, and provide appropriate initial reply information based on the understanding. The screening of the minimum answer level can improve the accuracy of the initial reply information and avoid the user's need to return to the previous answer level to operate. The query guidance is supplemented according to the judgment content, so that the displayed query guidance is more in line with the user's historical query situation.

[0052] By calculating the frequency of occurrence based on the proportion of occurrence, we can get the degree of influence of different keywords on different reply contents. By setting the modification coefficient, we can compare different differential reply contents and screen out differential reply contents that better meet the modification conditions. By setting the modification level, we can limit the scope of modification to avoid affecting the operation of the self-service answering system.

[0053] By comparing the modification level and the response level, the minimum response level that can replace the ideal response level can be accurately found. The setting of the information to be adjusted can divide the ideal response information into response levels according to the reference information, making the setting of the ideal response information more reasonable and reducing the impact caused by the addition of the ideal response information. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1This is a flow chart of the implementation method of the power IT service self-service response of the present invention. DETAILED DESCRIPTION

[0055] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0056] Example 1

[0057] like Figure 1 As shown, it is a flow chart of the implementation method of the self-service response of the power IT service of the present invention. The present invention provides an implementation method of the self-service response of the omni-channel power IT service, and the implementation method includes the following steps:

[0058] S100, establishing a response database, and classifying response levels according to response keywords;

[0059] S200, obtaining the user's query habits and query content through a questionnaire, and adjusting them to obtain adjusted query content;

[0060] S300, determining a response level for adjusting the query content, and providing the user with initial response information and query guidance according to the response level;

[0061] S400, tracking the user's subsequent operations and obtaining subsequent response information corresponding to the query guidance;

[0062] S500: When the number of guidance times reaches the guidance threshold, the manual service platform gives an ideal response information;

[0063] S600, the initial reply information and the subsequent reply information are recorded as reference information, and the manual service platform compares the ideal reply information with the reference information to obtain differential reply content;

[0064] S700, determining the difference keywords that cause the difference reply content, obtaining the response level and the frequency of occurrence of the difference keywords in the reference information, and determining the modification level of the difference reply content;

[0065] S800, modify the reference information according to the modification level to obtain modified reply information, replace the reference information with the modified reply information, and input it into the response database.

[0066] In step S100, a variety of reply information is stored in the response database, which is combined with different response keywords. Different response keywords have different response levels when combined, and different response levels will also provide different reply information. The lower the response level, the more general the explanation of each keyword in the reply information. The higher the response level, the more detailed the explanation of a certain response keyword.

[0067] In step S200, the self-service answering implementation method can be applied to APP or to the self-service equipment in the service hall. When each user creates an account, he / she reflects his / her query habits to the system by answering the system preset questionnaire. The query habits include but are not limited to query categories and language preferences. Language preferences are mainly to avoid users having specific spoken habits when using voice input to query questions, such as synonym replacement and sentence inversion. These habits may cause the answer database to misidentify. Therefore, the answer database needs to perform a preliminary translation of the initial query content according to the user's query habits.

[0068] In step S300, in some cases, the user will only enter keywords when making a query. When the number of keywords is small, the system cannot directly identify the user's true query intention and will use questions to narrow the user's query scope. The smaller the query scope of the query content is adjusted, the higher the corresponding response level will be. When there are relevant query records in the user's historical query records, the system will directly jump to the corresponding response level based on the information in the query record, so that the user does not need to click on the options in sequence according to the response level. The query guidance is a further explanation of the keywords in the initial reply information, which is convenient for users to understand the initial reply information, or the content that users may need to query based on the adjusted query content.

[0069] For example, the user's adjustment query content is broadband processing. The initial response information provided at this time is an introduction to broadband processing, and the query guide includes the time of broadband processing, the preparation materials for broadband processing, the cost of broadband processing, etc. If the user has previously searched for the keyword "preparation materials for broadband processing" based on historical query records, and the query content is whether a copy of the ID card is required for broadband processing, at this time, when the user inputs the adjustment query record for broadband processing, the initial response information is provided based on the key point of "preparation materials for broadband processing."

[0070] In step S400, the target time is usually 10 minutes, which can be adjusted according to the number of words in the initial reply information. Usually, the more words in the initial reply information, the longer the target time. The result of a single question and answer cannot determine whether the initial reply information meets the user's needs. Therefore, it is necessary to track the user's subsequent query content. At the same time, when the user frequently uses the query guidance, it means that the user is not satisfied with the content of the initial reply information. It is necessary to count such situations to facilitate the adjustment of the self-service answer method.

[0071] In step S500, the number threshold is related to the highest response level of the adjusted query content. Normally, when the value of the number of guidance times is greater than the value of the highest response level, the adjusted query content is transferred to the manual service platform. The ideal reply information is the reply information obtained by the user after consulting the staff. After the user gets a satisfactory reply, the staff will organize the reply information provided by themselves, obtain the ideal reply information, and upload it to the self-service answer system.

[0072] In step S600, when obtaining reference information, the user needs to record each subsequent reply information obtained according to the query guide, and the staff compares the reference information with the ideal reply information to obtain the difference reply content.

[0073] In step S700, the difference keywords are determined according to the difference reply content. When the difference keywords do not exist in the reference information, the frequency of occurrence of the difference keywords is 0. The staff needs to first determine the response level for the difference keywords. The obtained response level is the modification level of the difference reply content. When the difference keywords exist in the reference information, the modification level of the difference reply content is determined according to the frequency of occurrence of the difference keywords at each response level.

[0074] For example, an electric power company has launched a new service, but the response content of this service has not been stored in the self-service answering system, or the stored response content is not comprehensive enough. This will result in the self-service answering system being unable to obtain the desired answer based on the adjusted query content. After identifying this situation, the staff needs to set corresponding response levels for each different keyword to improve the convenience of users using the self-service answering system.

[0075] In step S800, the modified reply information is replaced to the specified position in the reference information according to the modification level. It can be understood that for some different keywords, the reference information and the modified reply information can be directly superimposed to facilitate users to better understand the modified reply information.

[0076] The setting of the response database allows the self-service response system to store enough response keywords. The division of response levels can provide corresponding reply information based on the user's query information. When the user enters fewer response keywords, reasonable query guidance can be provided to the user to improve the user's query experience. The acquisition of query habits can formulate different keyword reading methods according to the habits of different users, which improves the practicality of the self-service response method. By tracking the subsequent query content, it can determine the user's satisfaction with the initial reply information, and determine whether the initial reply information needs to be modified based on the satisfaction level. By setting the number of guidance times and guidance thresholds, the manual service platform can intervene in time when the self-service response system cannot provide satisfactory reply information. By obtaining the response level and frequency of occurrence of different keywords, it can determine the reference information that needs to be replaced for the different reply content. After the reference information is modified by the different reply content, the initial reply information corresponding to each response level is more reasonable.

[0077] Example 2

[0078] In a specific embodiment, the query habits of the user are obtained through a questionnaire, and the initial query content input by the user is adjusted according to the query habits to obtain the adjusted query content, which specifically includes:

[0079] S210, obtaining the word order habit of the user when performing a query through word order detection, and obtaining a word order adjustment template;

[0080] S220, obtaining the user's replacement habit for synonyms when performing a query through vocabulary detection, and obtaining a vocabulary adjustment template;

[0081] S230: Adjust the initial query content according to the vocabulary adjustment template and the word order adjustment template to obtain adjusted query content.

[0082] In step S210 to step S230, some users use voice input when inputting the initial query content, and different users speak in different ways. For example, some users like to use inversion when speaking, which will cause the order of keywords recognized by the self-service answering system to change. When the order changes, the initial reply information and query guidance may also change accordingly, causing inconvenience to the user. Similarly, different users use different words when describing the same thing. By acquiring replacement habits, synonyms that may be replaced are added to the answer database, which is convenient for the self-service answering system to acquire the initial query content.

[0083] Therefore, after investigating the behavioral habits of each user in the form of a questionnaire, a word order adjustment template and a vocabulary adjustment template can be obtained. The word order adjustment template and the vocabulary adjustment template can be continuously expanded during the use of the self-service answering system. The query habits of each new user can be determined through the word order adjustment template and the vocabulary adjustment template to determine the user category, and the initial query content provided by the user can be adjusted according to the user category to obtain adjusted query content.

[0084] Preferably, error messages for some commonly used answering keywords can also be stored, and when the corresponding error message is input, the self-service answering system directly translates the error message into the correct answering keyword.

[0085] The setting of the word order adjustment template can upgrade the self-service answering system according to the speech habits of different users, and can more accurately identify the initial query content entered by the user through voice input. The setting of the vocabulary adjustment template allows the self-service answering system to actively adjust the initial query content entered by the user according to the user's word usage habits, so that the self-service answering system can provide appropriate initial reply information.

[0086] Example 3

[0087] In a specific embodiment, obtaining the historical query records of the user, determining the response level of adjusting the query content according to the historical query records, and providing initial reply information and query guidance according to the response level specifically include:

[0088] S310, obtaining historical keywords of historical query records, matching the historical keywords with the response content in the response database, screening the response content according to the matching similarity, and obtaining the judgment content;

[0089] S320, screening out the minimum value of the response level corresponding to the determination content, and determining the response level of the adjusted query content according to the minimum value of the response level;

[0090] S330: Determine the detail of the response content according to the response level to obtain initial reply information, and supplement the query guidance according to the determined content.

[0091] In step S310, the historical keywords are matched with the response contents, the response contents are screened according to the matching similarity, and the response contents with matching similarity greater than the matching threshold are recorded as the determined contents.

[0092] In step S320, each judgment content has a corresponding response level, and the minimum response level is used as the basis. This can help the user eliminate some invalid options when the response keywords entered by the user are insufficient, thereby improving the user's query efficiency.

[0093] For example, the response level of the keyword input by the user is level 1, and the minimum response level of the judgment content is level 3. If no adjustment is made based on the judgment content, the response level of the query guidance obtained by the user is level 2. However, after adjustment based on the judgment content, the user can directly obtain a query guidance with a response level of level 4.

[0094] In step S330, after the response level is determined, the corresponding initial reply information is provided according to the response level. The content of the initial reply information is determined by the minimum value of the response level of the judgment content. If there are multiple judgment contents with the minimum value of the response level, the initial reply information corresponding to the multiple judgment contents are integrated. At the same time, the query guidance corresponding to the initial reply information should also be modified according to the judgment content. The relevant information in the judgment content will appear in the query guidance regardless of the response level. When the user selects the query guidance or the query guidance is not selected for three consecutive times, the relevant information of the judgment content will be deleted from the query guidance.

[0095] It should be noted that, normally, the query guide will only display the relevant information of the next response level, but when the query guide is newly added due to the determination content, there is no need to consider the response level corresponding to the determination content.

[0096] The acquisition of historical keywords can obtain the user's query content before using the self-service answering system, and can quickly understand the user's understanding of the answer keywords, and provide appropriate initial reply information based on the understanding. The screening of the minimum answer level can improve the accuracy of the initial reply information and avoid the user's need to return to the previous answer level to operate. The query guidance is supplemented according to the judgment content, so that the displayed query guidance is more in line with the user's historical query situation.

[0097] Example 4

[0098] In a specific embodiment, the initial reply information and the subsequent reply information obtained by the user are recorded as reference information, and the manual service platform compares the ideal reply information with the reference information to obtain the difference reply content, which specifically includes:

[0099] S610, summarizing and integrating the content of the ideal reply information to obtain ideal keywords;

[0100] S620, recording the initial reply information or subsequent reply information that does not contain the ideal keyword in the reference information as an erroneous reply content;

[0101] S630. Record the initial reply information or subsequent reply information containing the ideal keyword in the reference information as the differential reply content.

[0102] In step S610 to step S630, the ideal keywords are confirmed by the staff who provides the ideal answer information, and after confirmation, they are input into the self-service answer system. The self-service answer system screens the reference information according to the ideal keywords and obtains the wrong answer content and the difference answer content.

[0103] The setting of ideal keywords makes it more convenient to screen out incorrect reply contents and differential reply contents. The acquisition of differential reply contents can determine the specific location where the ideal reply information can be supplemented, making it easier for staff to carry out subsequent operations.

[0104] Example 5

[0105] In a specific embodiment, determining the difference keywords that cause the difference reply content, obtaining the response level and the occurrence frequency of the difference keywords in the reference information, and determining the modification level of the difference reply content specifically includes:

[0106] S710, obtaining the number of occurrences of the difference keywords in each difference reply content, and calculating the frequency of occurrence according to the number of occurrences;

[0107] S720, combining the response level and the occurrence frequency of each difference reply content to obtain a modification coefficient of each difference reply content;

[0108] S730. Determine the modification level of the difference reply content according to the modification coefficient.

[0109] In step S710 to step S730, the difference reply content contains the same ideal keywords as the ideal reply content, but the difference reply content cannot satisfy the user. The staff needs to analyze the difference reply content according to the adjusted query content provided by the user, and determine the reasons why the difference reply content is different from the ideal reply content, and analyze it in combination with the answer database to obtain the difference keywords. The difference keywords can be one or more, and each difference keyword has a corresponding number of occurrences and occurrence percentage. The number of occurrences is counted according to the number of difference keywords in the difference reply content, and the occurrence percentage is calculated according to the number of words in the difference reply content and the number of words in the difference keywords.

[0110] After calculation, the modification coefficient and modification level of each differential response content can be obtained. In order to avoid the ideal response content affecting the previous response system after modification, the modification level represents the lowest level that the differential response content can accept modification.

[0111] Obtaining the number of occurrences can obtain the occurrence ratio of different keywords. By calculating the occurrence frequency based on the occurrence ratio, we can obtain the influence of different keywords on different reply contents. By setting the modification coefficient, we can compare different different reply contents and screen out the different reply contents that better meet the modification conditions. By setting the modification level, we can limit the scope of modification to avoid affecting the operation of the self-service answering system.

[0112] Example 6

[0113] In a specific embodiment, the response level and the occurrence frequency of each difference reply content are combined to obtain the modification coefficient of each difference reply content, which specifically includes:

[0114] S721, calculating the difference percentage corresponding to each response level according to the occurrence frequency, and comparing the difference percentage with the difference threshold;

[0115] S722, when the difference percentage is greater than or equal to the difference threshold, adjusting the response level according to the difference percentage;

[0116] S723. When the difference percentage is less than the difference threshold, calculate a modification coefficient according to the difference percentage.

[0117] In step S721 to step S723, the number of words in the difference reply content is recorded as A1, the number of words in the difference keyword is recorded as A2, the number of occurrences of the difference keyword is recorded as B, the frequency of occurrence is recorded as Z, and the following relationship is satisfied among A1, A2, B and Z:

[0118] Z=A1÷(A2×B).

[0119] There are multiple differential response contents, and the response level corresponding to each differential response content is also different. Therefore, it is necessary to calculate the difference percentage corresponding to each response level.

[0120] For example, the number of different replies for response levels 3 to 6 is 4, 7, 6 and 3 respectively. At this time, the proportion of the number of response levels 3 to 6 is 0.2, 0.35, 0.3 and 0.15 respectively. The frequency of occurrence is recorded as the weight, and the difference percentage of each response level is calculated based on the proportion of the number.

[0121] After the difference percentage is calculated, it is compared with the difference threshold. The difference threshold is stored in the response database. When the initial response information is modified by the ideal response information, the initial response information with a smaller amount of change is given priority to facilitate the staff's operation. Therefore, for the difference response content with a large difference percentage, the response level needs to be converted according to the difference percentage to obtain the optimal modification level of the ideal response information.

[0122] When the difference percentage is less than the difference threshold, the best modification level for the ideal response information is preliminarily determined, and the modification coefficient is calculated based on the difference percentage.

[0123] Different response levels have different modification coefficients. The larger the response level, the larger the corresponding modification coefficient. Let the response level be L1, the modification coefficient be F1, and the difference percentage be F0. Then the following relationship is satisfied between F0, F1 and L1:

[0124] F1=1+F0×(0.8+0.4L1).

[0125] The calculation of the difference percentage can obtain the difficulty of combining the ideal response information with the initial response information of different response levels. The setting of the difference threshold can screen a large number of difference response contents and obtain the response level that is convenient for the staff to modify the response information.

[0126] Example 7

[0127] In a specific embodiment, the reference information is modified according to the modification level to obtain modification reply information, and the reference information is replaced with the modification reply information, specifically including:

[0128] S810, finding corresponding subsequent reply information from the reference information according to the modification level, and when the modification level is the same as the response level of the subsequent reply information, adding the ideal reply information to the subsequent reply information to obtain modified reply information;

[0129] S820, when the modification level is different from the response level of the subsequent reply information, calculating the level difference between the modification level and the response level, and recording the subsequent reply information between the level differences as information to be adjusted;

[0130] S830, integrating the information to be adjusted and the ideal response information to obtain multiple response vectors, performing similarity analysis on the multiple response vectors and the initial response information, and determining to modify the response information according to the analysis result.

[0131] In step S810, the modification level determines the minimum response level at which the ideal reply information can be replaced. When screening subsequent reply information, the response information with a smaller response level is preferentially selected for replacement. In this way, the user can obtain the relevant content of the ideal reply information using fewer query instructions and perform subsequent operations in a timely manner. For response levels lower than the modification level, the relevant content of the ideal reply information is not added to the corresponding subsequent reply information. This is to avoid the relevant content of the ideal reply information interfering with the user's query and affecting the query efficiency.

[0132] Therefore, after screening, the ideal reply information with the same modification level as the subsequent reply level is directly added to the subsequent reply information to obtain modified reply information.

[0133] In step S820, the range of subsequent reply information that can be modified by the ideal reply information is calculated through the level difference. For example, the modification level is 3 and the subsequent reply information is 7. The level difference at this time is 4, that is, the ideal reply information can be added to the subsequent reply content of response level 3 to response level 7. In order to allow users to obtain the relevant content of the ideal reply information using fewer query instructions, the modification position of the ideal reply information cannot be determined simply based on the response level of the subsequent reply information. The response database will divide the ideal reply information into response levels based on keywords, and form corresponding response keywords and put them into the subsequent reply information between the level differences. These subsequent reply information that can be placed in the ideal reply information are all recorded as information to be adjusted.

[0134] It should be noted that when calculating the level difference, the modification coefficients obtained in steps S721 to S723 need to be rounded up first.

[0135] In step S830, the subsequent reply information is vectorized by the BERT pre-trained language model to obtain a reply vector. Then all reply vectors are standardized to eliminate the influence of different vector magnitudes, and finally an answer vector is obtained at each response level.

[0136] Furthermore, a similarity analysis is performed between each answer vector and the initial reply information, and modified reply information is obtained according to the analysis result.

[0137] By comparing the modification level and the response level, the minimum response level that can replace the ideal response level can be accurately found. The setting of the information to be adjusted can divide the ideal response information into response levels according to the reference information, making the setting of the ideal response information more reasonable and reducing the impact caused by the addition of the ideal response information.

[0138] Example 8

[0139] In a specific embodiment, the information to be adjusted and the ideal response information are integrated to obtain multiple response vectors, the multiple response vectors are analyzed for similarity with the initial response information, and the response information is modified according to the analysis result, specifically including:

[0140] S831, filter out a reply vector containing ideal reply information and record it as a vector to be replaced;

[0141] S832: compare the similarities of the vectors to be replaced, use the vector to be replaced with the highest similarity as the modified reply information, and adjust the query guidance according to the order of similarity from high to low.

[0142] In step S831 to step S832, the reply vector with the highest similarity is used as the modified reply information. After the modified reply information is determined, the remaining reply vectors containing the ideal reply information are used as the vectors to be replaced. They are sorted according to their similarities, and the number of vectors with the same value as the level difference is selected from large to small. These vectors to be replaced are added to the query guidelines of each response level in order from large to small according to their similarity.

[0143] The selection of the vector to be replaced can further modify the query guidance of the modified reply information after the modified reply information is determined, so as to facilitate the user to search for the ideal keyword and make the query layout of the adjusted query content more reasonable.

[0144] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for implementing self-service response of omni-channel power IT services. It is characterized in that The following steps are involved: S100, establishing a response database, and classifying response levels according to response keywords; S200, obtaining the user's query habits and query content through a questionnaire, and adjusting them to obtain adjusted query content; S300, determining a response level for adjusting the query content, and providing the user with initial response information and query guidance according to the response level; S400, tracking the user's subsequent operations and obtaining subsequent response information corresponding to the query guidance; S500: When the number of guidance times reaches the guidance threshold, the manual service platform gives an ideal response information; S600, the initial reply information and the subsequent reply information are recorded as reference information, and the manual service platform compares the ideal reply information with the reference information to obtain differential reply content; S700, determining the difference keywords that cause the difference reply content, obtaining the response level and the frequency of occurrence of the difference keywords in the reference information, and determining the modification level of the difference reply content; S800, modify the reference information according to the modification level to obtain modified reply information, replace the reference information with the modified reply information, and input it into the response database.

2. According to claim 1, a method for implementing self-service response of omni-channel power IT services, It is characterized in that In step S200, the query content is adjusted, including the following steps: S210, performing word order detection on the questionnaire survey results, obtaining the word order habits of the user when performing a query, and obtaining a word order adjustment template; S220, obtaining the user's replacement habit for synonyms when performing a query through vocabulary detection, and obtaining a vocabulary adjustment template; S230: Adjust the initial query content according to the vocabulary adjustment template and the word order adjustment template to obtain adjusted query content.

3. According to claim 1, a method for implementing self-service response of omni-channel power IT services, It is characterized in that In step S300, the user's historical query records are obtained, the response level for adjusting the query content is determined according to the historical query records, and initial reply information and query guidance are provided according to the response level.

4. According to claim 3, a method for implementing omni-channel power IT service self-service response, It is characterized in that The following steps are involved: S310, obtaining historical keywords of the historical query record, matching the historical keywords with the response content in the response database, screening the response content according to the matching similarity, and obtaining the judgment content; S320, screening out the minimum value of the response level corresponding to the determination content, and determining the response level of the adjusted query content according to the minimum value of the response level; S330, determining the detail of the response content according to the response level to obtain initial reply information, and supplementing the query guidance according to the determined content.

5. According to claim 1, a method for implementing self-service response of omni-channel power IT services, It is characterized in that In step S600, obtaining the difference reply content includes the following steps: S610, summarizing and integrating the content of the ideal reply information to obtain ideal keywords; S620, recording the initial reply information or the subsequent reply information that does not contain the ideal keyword in the reference information as an erroneous reply content; S630: Record the initial reply information or the subsequent reply information containing the ideal keyword in the reference information as differential reply content.

6. A method for implementing self-service response of omni-channel power IT services according to claim 1, It is characterized in that In step S700, the difference reply contents are summarized and integrated to obtain difference keywords, the response level and the occurrence frequency of the difference keywords in the reference information are obtained, and the modification level of the difference reply contents is determined, including the following steps: S710, obtaining the number of occurrences of the difference keywords in each difference reply content, and calculating the frequency of occurrence of the difference reply content according to the number of occurrences of the difference keywords; S720, combining the occurrence frequency and the response level of each difference reply content to obtain a modification coefficient of each difference reply content; S730. Determine the modification level of the difference reply content according to the modification coefficient.

7. A method for implementing self-service response of omni-channel power IT services according to claim 6, It is characterized in that In step S720, the modification coefficient is obtained, including the following steps: S721, calculating the difference percentage corresponding to each response level according to the occurrence frequency, and comparing the difference percentage with the difference threshold; S722, when the difference percentage is greater than or equal to the difference threshold, adjusting the response level; S723. When the difference percentage is less than the difference threshold, calculate a modification coefficient according to the difference percentage.

8. A method for implementing self-service response of omni-channel power IT services according to claim 7, It is characterized in that In step S710, the number of words in the difference reply content is recorded as A 1 , the number of words of the different keywords is A 2 , the number of occurrences of the difference keyword is B, the frequency of occurrence is Z, and A 1 , A 2 , B and Z satisfy the following relationship: Z=A 1 ÷(A 2 ×B); In step S720, different response levels have different modification coefficients. The larger the response level, the larger the corresponding modification coefficient. The response level is L 1 , the modification factor is F 1 , the difference percentage is F 0 , then F 0 、F 1 With L 1 The following relationship is satisfied: F 1 =1+F 0 ×(0.8+0.4L 1 )。 9. A method for implementing self-service response of omni-channel power IT services according to claim 1, It is characterized in that In step S800, the following steps are included: S810, finding corresponding subsequent reply information from the reference information according to the modification level; when the modification level is the same as the response level of the subsequent reply information, adding the ideal reply information to the subsequent reply information to obtain modified reply information; S820, when the modification level is different from the response level of the subsequent reply information, calculating the level difference between the modification level and the response level, and recording the subsequent reply information between the level differences as information to be adjusted; S830, integrating the information to be adjusted and the ideal response information to obtain multiple response vectors, performing similarity analysis on the multiple response vectors and the initial response information, and determining to modify the response information according to the analysis result.

10. A method for implementing self-service response of omni-channel power IT services according to claim 9, It is characterized in that In step S830, the following steps are included: S831, filter out a reply vector containing ideal reply information and record it as a vector to be replaced; S832: Compare the similarities of the vectors to be replaced, use the vector to be replaced with the highest similarity as the modified reply information, and adjust the query guidance according to the order of similarity from high to low.

Images